Background Tumor microenvironment (TME)-targeting agents such as anti-angiogenic therapies and check-point inhibitors (CPIs), have shown both promise and variability in effectiveness depending on the tumor type. For immune-targeting agents like CPIs, efforts to identify features or biomarkers that predispose responding patients include but are not limited to genomic stability, tumor mutation burden, and PD-L1 expression. Oncologie is developing a RNA-based platform that identifies subsets of patients based on multiple aspects of the biological processes (dominant biology) existing within the tumor microenvironment.

Methods RNA data from publicly available sources including microarray, RNASeq exome and whole RNA were analyzed with respect to gene signatures that describe four different microenvironmental phenotypes. Phenotypes were then evaluated for relationships to clinical efficacy endpoints. From these RNA signatures and driven by machine learning methodologies, drug-specific algorithms were developed and applied to retrospectively to clinical data. Comparative analyses were explored between gene signatures, commonly used biomarkers (eg. presence of microsatellite DNA, expression levels of PD-L1, etc) and within-patient metadata to better understand better how this approach can be utilized in prospective clinical studies.

Results Attributes in RNA expression identified using Oncologie’s platform have retrospectively characterized responders to CPIs or anti-angiogenic drugs, demonstrating a relationship between clinical response and biomarker positive and negative patient populations. Exploratory data summarizing the use of the this platform demonstrates its utility for enriching response to both immune- and angiogenesis-targeting drugs. Relative expression changes between archival and fresh biopsies demonstrate changes in the TME with time and/or following targeted therapy. Lastly, cross-tumor comparisons support a tumor-agnostic utility of this approach. Detailed comparisons of this biomarker approach relative to other available biomarkers will be presented for standard of care drugs and those in the Oncologie pipeline based on retrospective analyses.

Conclusions RNA based descriptors of biology may be a useful approach to enrich for response to targeted therapies whose mechanism of action is to modify the TME biology.